I’ve referred a few people to a recent paper of ours regarding an analysis of Twitter. It’s worth mentioning a result not included in the paper. I think I briefly mentioned this issue in the past.
When our PNNL colleagues initially ran betweenness centrality on the feed related to the Atlanta flood, they looked at all posts and not just those that were in conversations. Because of “re-tweets” (re-sending the message with attribution) by public accounts, a few private accounts were identified as highly important.
It ends up that I know some of the people whose then-private accounts were considered important. And they were important. They provided very useful information, and the folks with public accounts did the right thing in one sense by providing full attribution.
Now there are articles about the UK protests (found thanks to Glyn Moody) mentioning that identifying leaders on Twitter is difficult. Don’t necessarily believe it, even if the analysis is restricted purely to public data. Similar analysis could tackle shared items on Facebook, although it appears more rare there. Correlating the two (and other public sources) should provide even more information on private accounts.
If you post useful or valuable information and people provide attribution, your private account will be slightly less private. Providing attribution is important to many people, so… You must give up any claim to being a source to keep privacy and anonymity. Not news to some, but most people don’t keep that in mind.